Give AI agents the Gateway actions Ignition's built-in OpenAPI does not cover.

Reports the embedded Jython version, Python 2 behavior, UTF-8 boundaries, numeric behavior, and Java availability.

Runs a fixed syntax and behavior matrix that separates valid Jython 2.7 code from Python 3-only forms.

Checks Python 2 standard-library module names and bounded Queue, URL, HTTP, ConfigParser, and SocketServer behavior.

Tests safe import fallbacks for missing modules, members, nested imports, and broad exception handling.

Checks explicit UTF-8 encoding and decoding, str/unicode mixing, joins, formatting, and invalid codecs.

Proves selected Python 3-only APIs are absent and records the bounded Python 2 alternatives.

Exercises Python and Java exception routing, mixed catches, finally blocks, and cleanup-failure masking.

Checks Python 2 built-ins, eager map/filter/zip behavior, map(None, ...), reduce, and module naming.

Compares the embedded standard-library inventory with the fixed standalone Jython allowlist without importing arbitrary modules.

Resolves a fixed set of Java classes through Ignition's class loaders and reports bounded version and loader evidence.

Checks Python callback adaptation, callback arity, Java char typing, and Java compiler availability.

Tests fixed syntax, XML support, and bounded import attempts for dataclasses, NumPy, and pandas.

Checks warnings, restored sys.argv state, protocol-2 pickle behavior, interactive compilation, and Java 17 imports.

Separates pure-Python and Java-library options from unsupported native-extension assumptions.

Exercises fixed and varargs Java overload resolution with embedded classes and no external fixture.

Checks selected Unicode 15.1 names, the data version, and invalid-name behavior without inferring Python 3 support.

Proves fixed sys attributes remain protected and verifies temporary runtime state is completely cleaned up.

Tests monotonic clocks, integer-millisecond validation, deadline rounding, clamping, and one-budget allocation.

Exercises Future completion, worker failure, timeout, cancellation, interruption, and executor cleanup.

Classifies write quality, readback, stale values, mismatches, partial failures, and fully verified success.

Validates a fixed bounded message schema, including exact keys, types, counts, duplicate paths, and finite numbers.

Checks begin, work, commit, rollback, close, cleanup failure, and commit-outcome-unknown ordering.

Exercises duplicate and conflict suppression, retry authorization, bounded attempts, completion, and unknown outcomes.

Tests ordered scope, availability, resource, authorization, completion, and Vision UI-thread decisions.

Checks epoch units, strict parsing, explicit time zones, DST gaps and overlaps, and elapsed versus calendar time.

Tests a bounded diagnostic schema, severity policy, correlation IDs, sensitive fields, lazy debug work, and logger failures.

Checks Gateway Tag Change versus Value Changed selection, initial execution, missed events, quality, timestamps, and project lifecycle.

Exercises duplicate suppression, stale observations, conflicts, overflow, quality recovery, checkpoints, and outcome-unknown handling.

Observes fixed Project Library generation and thread-safe state across a two-phase import protocol.

Tests failed initialization, missing dependencies, circular imports, acyclic imports, and successful caching.

Checks how held module, function, and object references behave when a module is replaced or removed.

Checks captured functions and bound methods when module functions or class members are removed.

Tests changed defaults, keyword names, required arguments, positional order, and captured callables across generations.

Tests migrations from scalar, list, or None returns to exact versioned result envelopes.

Exercises callback registration, generation conflicts, replacement, unregister, captured callbacks, and empty cleanup state.

Reports availability of fixed database, transaction, tag, async, Vision, and Perspective functions in Gateway scope.

Checks Dataset and PyDataset iteration, JSON boundaries, malformed input, date round trips, time zones, and message-function availability.

Reports only the availability of current and legacy secret-management function names without reading any secret value.

Performs bounded configuration discovery without returning configuration values or calling write operations.

Reads value, quality, and timestamp evidence from two fixed Perspective test tags plus a missing-path case.

Tests a fixed click-counter function against valid inputs and exact type and range failures.

Dry-runs or writes one fixed Boolean test tag, then verifies write quality and exact readback.

Dry-runs or requests allowlisted navigation for one verified llm-tools browser session and page.

Creates or updates one owned Perspective view shape with scan locking, atomic replacement, and label readback.

Creates or updates one owned /llm-tests route while preserving page configuration and verifying the route count.

Creates or updates one fixed Gateway project-script module and verifies the saved revision after a scan.

Invokes the fixed project module with bounded inputs and validates the exact returned envelope.

Reads up to 64 scoped qualified tag paths with value, quality, timestamp, type, and bounded nested serialization.

Browses a scoped base path with recursive and type filters, sorting, truncation, and continuation evidence.

Dry-runs or applies up to 32 scoped scalar writes, with before/after reads, quality, equality, and per-item results.

Queries bounded raw history for exact scoped tag paths over a window of no more than 24 hours.

Inspects, creates, or replaces one validated grouped-UDT resource with dry-run, hash preconditions, and readback requirements.

Reads scalar, array, Dataset, or Document values with explicit size, depth, type, quality, and truncation boundaries.

Dry-runs or writes one Document tag under timestamp and SHA-256 preconditions, then verifies its hash and readback.

Queries bounded current alarm rows by scoped tag prefixes, state, priority, and selected associated data.

Lists bounded shelved alarm sources with expiration, expired state, user, completeness, and truncation fields.

Dry-runs or acknowledges exact scoped event IDs, then verifies each retained event is acknowledged.

Dry-runs or shelves exact scoped alarm sources for a bounded duration and verifies the shelved-path readback.

Dry-runs or unshelves exact scoped alarm sources and verifies they are absent from the shelved-path readback.

Queries one named alarm journal for exact event IDs, scoped paths, selected data, and a window of no more than one hour.

Dry-runs or creates up to 100 bounded, side-effect-free expression tags inside one owned run folder.

Creates server-owned expression profiles for sibling and dynamic references, quality behavior, and fixed runScript exception controls.

Inspects, dry-runs, or deletes only an owned expression run folder and verifies its absence.

Sets up, triggers, inspects, and cleans up one fixed Event Stream expression fixture with bounded source values.

Exercises a fixed Perspective view, Named Queries, route, locale, selection, tag run, navigation, inspection, and cleanup workflow.

Sets up, starts, gates, retargets, inspects, and cleans up fixed SFC transition-expression cases.

Exercises one fixed parallel-cancel SFC through setup, tag-driven start, cancellation, inspection, and cleanup.

Runs nine allowlisted SFC parallel-expression cases through setup, start, trigger change, inspection, and cleanup.

Manages the fixed disposable Vision skill lab and its button, snapshot, window-state, runtime, Client, and Gateway language-boundary bridges.

Reads bounded registered Vision session rows for the expected project without returning client addresses.

Exercises localhost-only raw-socket encoding, timeout, error, and cleanup cases.

Exercises fixed localhost system.net.httpClient health and refused-connection cases separately from raw sockets.